A speech recognition device that performs speech recognition of a speech relating, for example, to music is popular (see Japanese Patent No. 3911178).
When performing speech recognition of a speech relating to music, the speech recognition device acquires music information from a device, for example, a USB (Universal Serial Bus) memory. The speech recognition device generates dictionary data by applying dictionary conversion to name data of names (artist names, album names, title names, playlist names, and so on) contained in the acquired music information and registers the generated dictionary data into a dictionary. The speech recognition device performs speech recognition using the dictionary data registered in the dictionary.
In the related art, however, generation of the dictionary data is started after all name data contained in the music information is acquired (after a full set of name data is acquired), that is, generation of the dictionary data is not started until all the name data contained in the music information is acquired. Hence, when the number of pieces of music information as subjects for which the dictionary data is to be generated is large and a data volume of the music information is increased, it takes a long time until generation of the dictionary data is started. Consequently, the related art has a problem that it takes a long time after the device is connected to the speech recognition device until speech recognition becomes available.
Also, in a case where the dictionary data was generated the last time a device was connected and the same device is connected again, when the music information stored in the device is changed (added or deleted) even slightly since the last time the device was connected, the dictionary data generated at the last connection is discarded and new dictionary data is generated from the start. Hence, the related art has a problem that even when only a slightest amount of music information is changed, speech recognition becomes unavailable over too long a period for a data volume of slightly changed music information.